Prof. Michael Lin

Antigen-stabilizable fluorescent nanobodies that become fluorescent upon binding intracellular targets enable background-free multicolor imaging and biosensing in living systems. @EinsteinMed @salkinstitute @animmerj nature.com/articles/s4159…

@adamjrubin It's nice to see follow-up work on the phase separation observed with LAT by Ron Vale's lab (cited in your paper and below). If I understand your Fig 4h correctly, the idea is that condensation does create a step function in effector activation? science.org/doi/10.1126/sc…

Please reach out if you want to chat about the experimental or computational approaches, disordered protein biology, or anything else, and please see our paper for the full story! (4/4)

Excited to share our paper out today in @ScienceMagazine! We directed high-content, single-cell genetic screens to ask how a disordered adapter protein orchestrates the complex process of T cell activation. (1/4) science.org/doi/10.1126/sc…

AI for biology is a lot harder than most pure tech people think. There are many bottlenecks, and high-quality data is the biggest one. If you actually wanna move the needle on AI/bio, then we need the following: - A major global project to expand high-quality biological data. - More foundation models for biology at the AlphaFold calibre. - A breakthrough in graph neural networks, as significant as transformers were for NLP. - New AI architectures that are first principles designed to natively work on bio problems. - A more intellectually honest and balanced research/business culture than the current hype game. - A major effort to identify, isolate, and exclude scientifically fraudulent papers and data points in biology. - Policy changes for more funding, cheaper electricity, targeted deregulation around the bioeconomy and use of anonymised data etc. - could be useful too.

Excited to share our discovery of a new programmable RNA-guided DNA-targeting system hiding inside bacteriophages that predates CRISPR. We call it VIPR (Viral Interference Programmable Repeat), and it uses an entirely new logic to find its targets. Thread + link below.

@michelle_monje Congrats, Michelle!

So grateful for the honor of becoming an #NASmember! #NAS163

I find LLMs very helpful for scientific writing. I do the legwork of planning out what to say, track down citations, sketch out the flow, feed all of this to the LLM to generate a draft, and the output is so awful it motivates me to write it out the right way in disgust

Development of a clinically viable MRGPRX4 inverse agonist for cholestatic itch treatment nature.com/articles/s4158…

❗️❗️❗️NEWS︱Nature Communications 👏👏👏Prof. Shuwen Chang’s Team has elucidated the mechanisms linking phase separation and synaptic plasticity in autism. BrainVTA website: brainvta.tech 👇MORE nature.com/articles/s4146…

Indeed if you like technology and progress you should hate AI or at least the LLM version of AI, because its answers are imprecise at best and confabulated at worst, and it takes enormous energy while also sucking up financial resources that would be better used elsewhere.

One of the most jarring things about current AI is its lack of introspection ability and metacognition. It doesn't know what it doesn't know, how it knows, or how it could find out. It's a one-way system.

Oxford AI professor Michael Wooldridge: "ChatGPT doesn't understand anything. It's essentially doing some fancy statistics."

Have you wondered what the wet lab success rates are for current AI-driven protein design models? Look no further! In our new open access review, @KevinKaichuang, @avapamini, @SarahAlamdari, and I report wet lab success rates for *over 200* different protein design tasks 🧬💻

Excited to share that NegBioDB has been accepted into @AnthropicAI 's AI for Science program! A database of biological negative results: drug discovery failures, trial terminations, and PPI data. About 63M records spanning multiple biomedical domains.

A great week wrapping up the 4th Stanford/UCSC Advanced Techniques in Neuroimaging Course! Excellent talks, exciting demos. Thanks to everyone who made it such a stimulating working. Looking forward to next year! @StanfordBrain @ucsc #Neuroimaging #Zuolab

This 2-hour Stanford lecture breaks down how models like ChatGPT and Claude are actually built, clearer than what many people in top AI roles ever get exposed to. Save this and set aside two hours today. It might end up being the most valuable thing you learn all week.

RosettaCommons released a new bootcamp series covering the modern protein design stack. Five takeaways worth bookmarking if you are building or managing a design pipeline.

Run an MD simulation of any protein in the AlphaFold Protein Structure Database using AF-CALVADOS Thanks to Sören von Bülow, AF-CALVADOS is now on Colab colab.research.google.com/github/KULL-Ce…

How much design and screening do you need? Amazon had to design 288,000 nanobody constructs spanning eight target epitope regions and screened 100,000 using yeast display. Thats the scale still required for de novo design amazon.science/publications/a…

REMOTE CONTROL MICE! A group from Seoul has just published in Cell that they’ve discovered a crazy-weird protein (Cyb5b) that can be used to turn on any gene when exposed to EMFs! They used it to turn on OSK and extend the lifespan of a progeroid mouse. More to come on this…